Enhancing Performances of Perovskite Solar Cells by Using Zinc Oxide Quantum Dots/Mesoporous Titania Electron Transport Layer

As the photovoltaic performance of mesoscopic perovskite solar cells (PSCs) is strongly dependent on the interface between perovskite and electron transport layer. Herein, an attempt to load facile processed ZnO quantum dots (ZQDs) or TiO 2 quantum dots (TQDs) into the mp-TiO 2 layer via a simple spin-coating method was first performed. Both of them had huge impact on the morphology of perovskite films, leading to larger perovskite grains. However, the power conversion efficiency (PCE) of these two kinds of QDs modified cells exhibited a different tendency. The champion PCE of ZQDs modified PSCs was remarkably improved from 14.54% to 17.2%, while that of TQDs modified ones decreased to 11.78%. We demonstrated that the enhanced PCE and short-circuit current (J sc ) were attributed to the enlarged grain size and enhanced light absorption of perovskite film, faster electron extraction and transport as well as less recombination for ZQDs modified PSCs, which mainly resulted from the increased active specific surface area. On the contrary, deterioration of TQDs modified PSCs was exactly ascribed to the aggregation of TQDs which sharply decreased the specific surface area of the electron transport layer. The current work provided an efficient and facile way to improve the photovoltaic performance of mesoscopic PSCs.


Introduction
Since the organometal trihalide perovskite were reported as photosensitizers for photovoltaic cells, 1 they have attracted great attention for their low cost, simple manufacturing process and high photovoltaic performance in the field of optoelectronics, especially in the application of perovskite solar cells (PSCs). 2 The energy conversion efficiency of PSCs has risen from 3 % 1 to 25.5 % 3 in just a few years.
Mesoporous structured PSCs, especially those based on mesoporous TiO2 (mp-TiO2), have been proved to have the highest power conversion efficiency (PCE).
Because the mp-TiO2 has a large specific surface area, it can be used as a scaffold to increase the loading amount of perovskite. In addition, it is in direct contact with perovskite and acts as an electron transport layer (ETL), collecting photoelectrons from perovskite and transmitting them to the front contact. 4,5,6 Therefore, it will be a possible way to promote the performance of PSCs by increasing the effective specific surface area between the carrier transport layer and the light absorbing layer. 7,8,9,10,11 Due to the fact that quantum dots (QDs) are only a few nanometers in diameter, they are endowed with large specific surface area. Additionally, due to the influence of quantum confinement effect and edge effect, the band gaps of QDs are adjustable. These characteristics make them possess potential application prospect in the field of PSCs.
For instance, in planar heterojunction PSCs, black phosphorus QDs have been used to enhance the hole extraction. 12 CdSe QDs, graphene QDs and Carbon QDs have respectively been used to modify the ETL to make the perovskite thin films denser or accelerate the extraction rate of electrons. [13][14][15] According to this idea, selecting appropriate quantum dots to modify the mp-TiO2 is expected to obtain an idea carrier transport material with large specific surface area, appropriate energy level and high conductivity.
In this thesis, we plan to prepare ZnO QDs (ZQDs) and TiO2 QDs (TQDs), embedding them into mesoporous scaffolds to increase the specific surface area of mesoporous layer, and finally enhance the light absorption of perovskite layer and accelerate the extraction and transmission of electrons. On the one hand, ZnO and TiO2 have ever been employed as excellent ETL in the area of PSCs owing to their suitable energy levels. On the other hand, ZQDs and TQDs are easily synthesized, and they can by deposited into mesoporous layer by simple spin coating technique. In addition, ZnO has a high electron mobility of 115-155cm 2 V -1 s -1 , 16,17 which is favorable for electron transport.
With this strategy, ZQDs and TQDs modified mesoscopic PSCs have been prepared. Unexpectedly, the ZQDs modified mesoscopic PSCs showed an enhancement in PCE (PCE=17.20%) as compared to unmodified PSCs (PCE=14.54%), while TQDs modified ones showed an opposite tendency (PCE=11.78%). Further discussions indicated that enhanced light absorption, accelerated electron transport, extraction and less recombination were existed in ZQDs modified PSCs for their increased surface area. On the contrary, because of the aggregation of TQDs on the mp-TiO2, the pore volume and specific surface area were sharply decreased which ultimately led to the performance deterioration of the TQDs modified PSCs.
The fabrication process of PSC modified by QDs is shown in Figure 1.    To verify whether the QDs can work well in the devices, the surface energies were characterized by UV-vis absorption spectrum and Ultraviolet photoelectron spectroscopy (UPS) ( Figure S1). According to the absorption spectra, the band gaps   The MAPbI3 films were formed by spin coating the perovskite precursor solution on the bare or QDs modified mp-TiO2 substrate and then heating at 100℃ for 2 min. Figure 4,a b, c display the field emission scanning electron microscope (FESEM) topview images of the perovskite films on the bl-TiO2/mp-TiO2, bl-TiO2/mp-TiO2/ZQDs and bl-TiO2/mp-TiO2/TQDs substrates, respectively. We extracted the statistical distribution diagrams of perovskite grain size by analyzing the top-view SEM images using the software named nano measurer. From the inset of the SEM images, the grain size is in Gaussian distribution, the average size is 254nm, 420nm and 430nm for perovskite grown on the bl-TiO2/mp-TiO2, bl-TiO2/mp-TiO2/ZQDs and bl-TiO2/mp-TiO2/TQDs substrates, respectively. It shows that the perovskite grain size on the QDs modified substrates is obviously larger than that on the counterpart. The enlargement of perovskite grain sizes may be ascribed to the flowing reasons: First, heterogeneous nucleation is suppressed and less drag force is performed from the QDs modified substrates, yielding increased nucleus spacing and promoted grain boundary migration in grain growth. 18 Second, the QDs with high surface-to-volume ratio increase the contact area of perovskite/ETL, which is favorable for the formation of a high-quality film. 9 Third, QDs offer added nucleation sites for perovskite, thus larger grain size can be obtained. 12 Considering that the grain boundaries and the defect density of the perovskite film can be reduced due to the enlarged grain size, leading to decreased nonradiative recombination and increased charge transport, 19,20,21 this strategy would be beneficial for achieving high performance optoelectronic devices.
The XRD patterns of perovskite films on different mesoscopic layers are depicted in Figure 5. The stronger peaks of perovskite films coated on the QDs modified mp-TiO2 layers indicate higher crystallinity, in line with the results taken from the SEM images above. Therefore, enhanced performances are expected for these QDs modified PSCs.
Then the subsequent HTM and Au electrode were spin coated and thermally evaporated on the perovskite layers, respectively, to form the complete PSC devices.
The device with bare mp-TiO2 was marked as PT; the devices with ZQDs and TQDs modified mp-TiO2 were marked as PZT and PTT, respectively. Cross-sectional SEM images show in Figure 4d, e, f indicate that the thickness of MAPbI3 over ETLs is not changed significantly through QDs-treatment.    8.57 Figure 7a shows the current density-voltage (J-V) curves of the champion PSCs.
The photovoltaic parameters are recorded in Table 1. It shows that the champion PZT cell achieved a much higher PCE than that of the PTT cell and reference PT cell.
Moreover, the PZT cells showed a smaller J-V hysteresis, which may resulted from the improved electron injection/transport to balance the hole flux at the anode. 22 Figure 7b represented the steady current density at maximum power point as a function of time. Moreover, the effect of mp-TiO2 layer's thickness on the device efficiency was investigated. The thickness of mp-TiO2 layers was controlled by changing the spincoating speed from 2500 to 5000 rpm while keeping otherwise identical conditions. We found that the optimal spin-coating speed was 3000 rpm and the highest PCE values of the devices followed the sequence as PCE (PTT) < PCE (PT) < PCE (PZT) as before ( Figure 7d).
Seen from Table 1, the value of Voc was slightly changed, the change of PCE was mainly caused by Jsc. As a preliminary, we carried out the optical absorption studies.
For PSCs, the strong capacity of light absorption plays an important role in high device performance. Seen from the results in Figure 8, the MAPbI3 films exhibited a wide light absorbance range from the UV region to 800 nm. The MAPbI3 film grown on the ZQDs modified mp-TiO2 film (black curve) showed a considerable enhancement in UVvisible absorbance than the reference one (red curve), while the TQDs modified film (blue curve) showed a weakened light absorption.  The enhanced light absorption can help to explain the higher photocurrent and higher light harvesting efficiency in J-V curves above. 25 As no visible morphological difference was found between ZQDs-treatment and TQDs-treatment in the SEM top and cross-sectional images, the decreased light absorption for PTT film in Figure 8 (blue curve) may be ascribed to the reduced perovskite amount within the mesoporous layer caused by the blockage of pores.
Therefore, BET measurement was performed to investigate the pore volume and the specific surface area of the three samples. Figure 9d, e shows the N2 adsorption−desorption isotherms and the corresponding BET surface area plots of mp TiO2, ZQDs-mp TiO2 and TQDs-mp TiO2. The type IV (Brunauer -Deming Deming-Teller (BDDT) classification) isotherms for the above three samples indicated that mesopores were present. The type H3 hysteresis loops demonstrated the formation of slit-like pores, which were formed from the aggregation of the nanoparticles. 26 As the pore volume directly determines the amount of MAPbI3 within the mesoporous layer, the sharply decreased pore volume for the TQDs treated film will consequentially lead to the decreased fillers. Furthermore, the increased specific surface area of the oxides modified with ZQDs contributed to effective contact between MAPbI3 and ETL, which is a significant player in the improved photovoltaic performance (Table 2).  In addition to the enhanced light absorption, we suspected that the ZQDs-treated mp-TiO2 could improve the charge separation and extraction due to the increased contact interfaces. So, we conducted the steady-state photoluminescence (PL) and timeresolved PL (TRPL) measurements on the MAPbI3 films with different mesoporous layer, as shown in Figure 10. All samples exhibited a PL peak at 770 nm (excitation wavelength 532 nm) derived from the perovskite, which was in agreement with the band edge emission from MAPbI3 reported before. 27 Nevertheless, there existed great difference for the perovskite PL intensity among them. The PZT film exhibited a much lower PL intensity than PT film, suggesting a less radiative recombination which may resulted from the increased electron extraction from MAPbI3 to the electrode. Instead, the PTT film showed a much stronger PL intensity. To further study the difference of the behavior of excited electrons in the perovskite layer, TRPL measurements were performed. As displayed in Figure 10b, the average lifetimes of PT, PZT and PTT film were 21 ns, 14 ns and 28 ns, respectively, which were obtained by fitting the decay curves with a double-exponential function. 28 That is, the PZT sample exhibited the fasted PL quenching rate, which meant faster charge-carrier extraction across interface. 29,30 These results confirmed that the electron extraction was significantly speeded up by introducing ZQDs and the result was exactly opposite after replacing them with TQDs.
To further confirm the charge transfer at the perovskite / ETL interface, Nyquist plots in Figure 10c were recorded in dark over a frequency of 100 mHz to 1MHz under an applied bias of 0.7V. Seen from the Nyquist plots, a semicircle and an incomplete semicircle could be distinguished, which were corresponding to the charge transfer resistance (Rtr) and the recombination resistance (Rrec), respectively. 31 Here we mainly focus on the Rtr by fitting the first semicircle at high frequency region using the equivalent circuit model (inset

Conflicts of interest
There is no conflict to declare. Figure 1 Fabrication process of QDs modi ed PSCs.